CN105389574B - The method and system of human eye iris in a kind of detection picture - Google Patents
The method and system of human eye iris in a kind of detection picture Download PDFInfo
- Publication number
- CN105389574B CN105389574B CN201510984804.4A CN201510984804A CN105389574B CN 105389574 B CN105389574 B CN 105389574B CN 201510984804 A CN201510984804 A CN 201510984804A CN 105389574 B CN105389574 B CN 105389574B
- Authority
- CN
- China
- Prior art keywords
- center
- circle
- radius
- iris
- image
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/193—Preprocessing; Feature extraction
Abstract
The invention belongs to fields of communication technology, disclose a kind of method and system for detecting human eye iris in picture, including step obtains eyes image using client camera；Gray level image is converted by the eyes image；Calculating and rectangular block gray scale Data-Statistics are filtered to the gray level image, obtain the estimation center of circle and the estimation radius of iris；According to the estimation center of circle and estimation radius, is iteratively solved using the method in the progressive step-length resetting center of circle, obtain the accurate center of circle and the precise radius of iris；The accurate center of circle of iris and precise radius are sent to client.The present invention can be accurately positioned the center location and radius of iris, can preferably exclude the interference such as eyelashes and eyelid, greatly accelerate detection speed.
Description
Technical field
The invention belongs to human eye detection technical field, method more particularly to human eye iris in a kind of detection picture and it is
System.
Background technique
In recent years, with the progress of mobile phone front camera and rear camera, captured image is more and more clear, from
And the eye image obtained is also become better and better, research staff can obtain eye image by terminal and carry out many work: from peace
Present unlocked by fingerprint application is wider for full angle, but more much higher than the safety coefficient of fingerprint by iris unlock；Make from simplification
The angle being accustomed to mobile phone can also write an APP, control mobile phone page turning by the pupil moving direction in iris come first
Deng；For the angle of beautification photo, iris photo can be beautified, achieve the effect that wear U.S. pupil, or increase eyes to seem
More there is mind.And above series of operation has a premise, that is, first have to the center of circle and half that accurate detection goes out iris
Diameter.
The method of detection iris or pupil has the circumference Hough of circumference calculus of finite differences and Wildes based on Daugman to become at present
Change detection.Above method, it is higher to be required to sample photographs pixel, needs special shooting together；It cannot exclude eyelashes and eyelid
Deng interference；And traditional search center of circle method, search range is very big, needs to each point in image, calculating is with the point
A series of circumference of different radiis at center, then counts the score of each radius circumference, and detection speed is slower.
Summary of the invention
It to solve the above-mentioned problems, can the invention proposes a kind of method and system of human eye iris in detection picture
It is accurately positioned the center location and radius of iris, the interference such as eyelashes and eyelid can be preferably excluded, greatly accelerate inspection
Degree of testing the speed.
In order to achieve the above objectives, the technical solution adopted by the present invention is that: a method of detection picture in human eye iris, packet
Include step:
(1) eyes image is obtained using client camera；(2) gray level image is converted by the eyes image；(3) right
The gray level image is filtered calculating and rectangular block gray scale Data-Statistics, obtains the estimation center of circle and the estimation radius of iris；(4) root
According to the estimation center of circle and estimation radius, is iteratively solved using the method in the progressive step-length resetting center of circle, obtain the accurate circle of iris
The heart and precise radius；(5) the accurate center of circle of iris and precise radius are sent to client.
Further, the acquisition methods of the eyes image are to detect people by machine learning in the step (1)
Face obtains the key point of detection face, extracts eye key point and constitutes eyes image；Or it by user's interaction, clicks screen and obtains
Obtain eyes image.
Further, the eyes image is RGB image in the step (2), R, G and B in RGB image are utilized
Eyes image is converted gray level image by the value in three channels.
Further, the filtering calculates comprising steps of setting one in the gray level image in the step (3)
Centered on some pixel in row pixel；Find out the pixel near the central horizontal direction；Calculate the center and described attached
The gray value of nearly pixel；By the ascending sequence of gray value；The gray scale at the center is replaced with maximum or secondary big gray value
Value.By filtering the influence for calculating and capable of effectively reducing eyelash and informer etc. and detecting to Iris Location.
Further, the rectangular block gray scale Data-Statistics are comprising steps of in the gray level image in the step (3)
Middle creation horizontal rectangular block；The horizontal rectangular block is from left to right moved, when every one pixel of movement in statistics horizontal rectangular block
The sum of all pixels gray value and horizontal rectangular block center；The smallest horizontal rectangular block of the sum of gray value is found out,
Using its center as the horizontal position of pupil；Vertical rectangular block is created, by the vertical rectangular block in the horizontal position
Vertical sliding in the pixel of that column, when every one pixel of movement count the sum of all pixels gray value in vertical rectangular block with
And the center of vertical rectangular block；The smallest vertical rectangular block of the sum of gray value is found out, using its center as pupil
Vertical position；The gray value for counting that column pixel of the horizontal position forms grey level histogram；By the grey level histogram
The corresponding gray value in the lowest point obtains comparison threshold value multiplied by pre-determined factor；It is small to count gray value in that column pixel of the horizontal position
In the pixel of the comparison threshold value；By the number of institute's statistical pixel divided by 2, the pupil radium estimated；The water of the pupil
Prosposition sets the estimation center of circle with vertical position as iris, estimation radius of the pupil radium as iris.
Further, the horizontal rectangular frame height degree is eyes image height, width is four points of eyes image width
One of；The vertical rectangle frame height is a quarter of eyes image height, and width is eyes image width；The coefficient takes
Value is 1.4.
Further, in the step (4), the method iterative solution in the progressive step-length resetting center of circle, comprising steps of first will
The estimation center of circle finds out the scoring and respective radius in the initial center of circle as the initial center of circle；Then with the initial circle
Centered on the heart, according to the scoring and respective radius of progressive step length searching series of points around it；If some point is scored above just
The scoring in the beginning center of circle, just using the point as new center；The still series of points around it with progressive step length searching to new center
Scoring and respective radius, carry out the iteration of a new round；It, will if the scoring all put is respectively less than the scoring at center after iteration
The center is as the accurate center of circle, and using the respective radius at the center as precise radius.Further, the methods of marking
Comprising steps of setting initial radium, the sum of the gray value on part-circular periphery is counted according to the center；Later by the gray value
The sum of divided by the number of pixels on part-circular periphery, obtain circumference average gray value；Certain Radius is incremented to since initial radium,
Form radii sequence；The corresponding circumference average gray value of all radiuses in radii sequence is sought, average gray value sequence is obtained；By institute
It states the latter value in average gray value sequence and subtracts previous value, obtain new sequence, each value, all generation in the new sequence
The score of table respective radius；Find out scoring of the highest score in the new sequence as the center, highest score corresponding half
Best respective radius of the diameter as the center.Part-circular periphery is only considered in the scoring for calculating each center of circle, can be excluded very well
The case where eyelash blocks iris, and eyelid blocks iris.
Further, to center in the method for progressive step length searching scoring of series of points around it comprising steps of
(a) center scoring is calculated, as initial score；
(b) eight neighborhood for being separated by 2 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue；
(c) eight neighborhood for being separated by 4 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue；
(d) eight neighborhood for being separated by 6 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue；
(e) eight neighborhood for being separated by 8 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue；
(f) if the scoring of above-mentioned all the points is respectively less than initial score, then the center that iteration is gone out is the accurate center of circle, and
And using the optimum radius at the center as precise radius.
On the other hand, the present invention also provides it is a kind of detection picture in human eye iris system, including image collection module,
Image conversion module estimates computing module, accurately calculates module and output module；
Described image obtains module, obtains eyes image using client camera；
The eyes image is converted gray level image by described image conversion module；
The estimation computing module is filtered calculating and rectangular block gray scale Data-Statistics to the gray level image, obtains rainbow
The estimation center of circle of film and estimation radius；
It is described to accurately calculate module, according to the estimation center of circle and estimation radius, utilize the side in the progressive step-length resetting center of circle
Method iteratively solves the accurate center of circle and the precise radius of iris；
The accurate center of circle of iris and precise radius are sent to client by the output module.
Using the technical program the utility model has the advantages that
The method of human eye iris in a kind of detection picture proposed by the invention by first filtering, then counts rectangular block
The mode of gray value come calculate iris the estimation center of circle and estimation radius, can preferably exclude eyelashes etc. and be brought to iris detection
Interference effect；When being iteratively solved using the method in the progressive step-length resetting center of circle, by increasing at a distance from central point, change
In generation, searches for and calculates scoring, and all appearance more figures of merit are then reinitialized with the value, greatly accelerate detection speed.Institute of the present invention
It proposes a kind of system for detecting human eye iris in picture, method proposed by the invention can be cooperated to realize the application of this method.
Detailed description of the invention
Fig. 1 is the method flow diagram of human eye iris in a kind of detection picture of the invention；
Fig. 2 is eye image tonal gradation schematic diagram in the embodiment of the present invention；
Fig. 3 is the flow chart that calculating is filtered in the embodiment of the present invention；
Fig. 4 is the flow chart of rectangular block gray scale Data-Statistics in the embodiment of the present invention；
Fig. 5 is the flow chart for the method iterative solution that progressive step-length resets the center of circle in the embodiment of the present invention；
Fig. 6 is the flow chart of methods of marking in the embodiment of the present invention；
Fig. 7 is the methods of marking flow chart of multiple progressive step iteration search of eight neighborhood in the embodiment of the present invention；
Fig. 8 is the sequential schematic that iris is accurately positioned iterative search in the embodiment of the present invention；
Fig. 9 is the structural schematic diagram of human eye iris system in a kind of detection picture of the invention.
Specific embodiment
To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention is made into one with reference to the accompanying drawing
Step illustrates.
In the present embodiment, shown in Figure 1, the invention proposes a kind of method of human eye iris in detection picture, packets
Include step:
(1) eyes image is obtained using client camera.
The acquisition methods of the eyes image are to detect the key point that face obtains detection face by machine learning, mention
Eye key point is taken to constitute eyes image.
The acquisition methods of the eyes image can also be, by user's interaction, to click screen and obtain eyes image.
(2) gray level image is converted by the eyes image.
The eyes image is RGB image, is converted eyes image using the value in tri- channels R, G and B in RGB image
For gray level image.
Conversion formula is as follows:
Y=0.299 × R+0.587 × G+0.114 × B
Wherein, R, G, B are three channel values of color space in RGB image respectively.
(3) calculating and rectangular block gray scale Data-Statistics are filtered to the gray level image, obtain iris the estimation center of circle and
Estimate radius.
The position of pupil in eyes image, iris and sclera and the gray level image of corresponding position are shown in such as Fig. 2
Grade.By image it can be seen that pupil position is part most black in eye image, iris takes second place, and sclera is most bright.
Because pupil is most dark in image, it is possible to find initial circle of the most dark center as pupil in gray level image
The heart, if directly adopt gray scale and value it is minimum at ordinate and abscissa of the row and column as the initial center of circle, often by
The influence of informer's perhaps eyelash because eyelash or the gray value of informer are lower, come out in this way it is initial in
The heart is often biased to eyelash or informer.Therefore, rainbow is being obtained using filtering calculating and rectangular block gray scale Data-Statistics in the present invention
The influence of eyelash and informer is also reduced while the estimation center of circle of film and estimation radius.
(3.1) filtering calculates, as shown in Figure 3 comprising steps of setting in one-row pixels in the gray level image
Centered on some pixel；Find out the pixel near the central horizontal direction；Calculate the ash at the center and the pixel nearby
Angle value；By the ascending sequence of gray value；The gray value at the center is replaced with maximum or secondary big gray value.Pass through filtering
Calculating can effectively reduce the influence that eyelash and informer etc. detect Iris Location.
Specific implementation are as follows: centered on some pixel in setting a line.
Pixel near horizontal direction is respectively as follows:
pixeln-3,pixeln-2,pixeln-1,pixeln,pixeln+1,pixeln+2,pixeln+3
Corresponding gray value is respectively as follows:
Greyn-3,Greyn-2,Greyn-1,Greyn,Greyn+1,Greyn+2,Greyn+3
By the ascending sequence of these gray values, with maximum or secondary big gray value, the gray value replaced.
(3.2) the rectangular block gray scale Data-Statistics, as shown in Figure 4 comprising steps of being created in the gray level image horizontal
Rectangular block；From left to right move the horizontal rectangular block, all pixels when every one pixel of movement in statistics horizontal rectangular block
The center of the sum of gray value and horizontal rectangular block；The smallest horizontal rectangular block of the sum of gray value is found out, by its centre bit
Set the horizontal position as pupil；Vertical rectangular block is created, by the vertical rectangular block picture that is arranged in the horizontal position
Vertical sliding on element, when every one pixel of movement, count the sum of all pixels gray value in vertical rectangular block and vertical rectangle
The center of block；The smallest vertical rectangular block of the sum of gray value is found out, using its center as the vertical position of pupil；System
The gray value for counting that column pixel of the horizontal position forms grey level histogram；By the corresponding ash in the lowest point of the grey level histogram
Angle value obtains comparison threshold value multiplied by pre-determined factor；It counts gray value in that column pixel of the horizontal position and is less than the contrast threshold
The pixel of value；By the number of institute's statistical pixel divided by 2, the pupil radium estimated；The horizontal position of the pupil and vertical
The estimation center of circle of the position as iris, estimation radius of the pupil radium as iris.
It is embodied as, creates a horizontal rectangular block, the height RectH=of the horizontal rectangular block in the picture first
LImageHeight, width RectW=lImageWidth/4；Wherein, lImageHeight is the height of eyes image,
LImageWidth is the width of eyes image；From left to right mobile and horizontal rectangular block when every one pixel of movement, counts rectangle
The sum of all pixels gray value in block, GreySum are the sum of grey scale pixel value in horizontal rectangular block, and grey is level matrix block
The gray value of the interior each position of correspondence.Meanwhile the center RectCenterX of recording level rectangular block；From left to right traverse
After primary, the smallest horizontal rectangular block GreySum_min of the sum of gray value and its corresponding center are found out
RectCenterX_min.Tentatively assert that this center RectCenterX_min is the horizontal position ectCenterX of pupil.
Secondly, vertical rectangular block is created, the height RectH=lImageHeight/4 of the vertical rectangular block, width RectW
=lImageWidth；Wherein, lImageHeight is the height of eyes image, and lImageWidth is the width of eyes image；
By vertical sliding in the vertical rectangular block pixel that is arranged in the horizontal position, with the horizontal position class with above-mentioned pupil
The vertical position estCenterY of pupil is found out like method.
Then, the gray value for counting that column pixel of horizontal position estCenterX forms grey level histogram, most black
That a part be grey level histogram the lowest point belong to pupil certainly.Assuming that the corresponding gray value in grey level histogram the lowest point is
GreySum_min, to this lowest point minimum value multiplied by pre-determined factor, pre-determined factor is set as 1.4 in the present invention；By 1.4 ×
GreySum_min threshold value as a comparison；It counts gray value in that column pixel of the horizontal position and is less than the comparison threshold value
Pixel all takes pupil pixel as；By the number of institute's statistical pixel divided by 2, the pupil radium estRadius that is estimated.
Finally, the estimated location (ectCenterX, ectCenterY) of pupil and the estimation radius of pupil
estRadius。
In the image of actual photographed, for Asian, pupil is usually black, and iris would generally brown or class
As it is dark, this is resulted in the picture, and pupil and iris are very close, so for the ethnic group of Asia, obtained pupil
Center and radius are center and the radius of iris.
Therefore, the horizontal position of the pupil and vertical position as iris the estimation center of circle (ectCenterX,
EctCenterY), estimation radius estRadius of the pupil radium as iris.
(4) it according to the estimation center of circle and estimation radius, is iteratively solved, is obtained using the method in the progressive step-length resetting center of circle
The accurate center of circle of iris and precise radius.
In the step (4), the method iterative solution in the progressive step-length resetting center of circle, as shown in Figure 5 comprising steps of first by institute
The estimation center of circle is stated as the initial center of circle, and finds out the scoring and respective radius in the initial center of circle；Then with the initial center of circle
Centered on, according to the scoring and respective radius of progressive step length searching series of points around it；If being scored above for some point is initial
The scoring in the center of circle, just using the point as new center；The still series of points around it with progressive step length searching to new center
Scoring and respective radius, carry out the iteration of a new round；It, should if the scoring all put is respectively less than the scoring at center after iteration
Center is as the accurate center of circle, and using the respective radius at the center as precise radius.
(4.1) methods of marking, as shown in Figure 6 comprising steps of setting initial radium, counts part according to the center
The sum of gray value on circumference；It is average to be obtained into circumference divided by the number of pixels on part-circular periphery for the sum of described gray value later
Gray value；It is incremented to certain Radius since initial radium, forms radii sequence；Seek the corresponding circle of all radiuses in radii sequence
All average gray values obtain average gray value sequence；Latter value in the average gray value sequence is subtracted into previous value,
New sequence is obtained, each value in the new sequence all represents the score of respective radius；Find out the best result in the new sequence
Scoring of the number as the center, best respective radius of the corresponding radius of highest score as the center.Calculating each circle
The case where scoring of the heart only considers part-circular periphery, can exclude eyelash very well and block iris, and eyelid blocks iris.
It is embodied as, is first with certain length R1 first centered on the estimation center of circle (ectCenterX, ectCenterY)
The reason of beginning radius counts the sum of the gray value on part-circular periphery GreyCircleSum1, takes part-circular periphery is iris often quilt
Upper lower eyelid blocks, and can effectively avoid influence of the eyelid to testing result.GreyCircleSum1 will be divided by circumference later
Pixel number, obtain circumference average gray value, specific formula is as follows:
CircleAvg1=GreyCircleSum1/num_Circle
Wherein num_Circle is the number of pixels on part-circular periphery.
Secondly, being incremented to some radius Rn since radius R1 with similar method, a radii sequence is obtained:
(R1, R2 ... Rn), wherein R1, which may be set to 5, Rn, may be set to 40.It is flat to find out the corresponding circumference of all radiuses in radii sequence
Equal gray value constitutes average gray value sequence:
(CircleAvg1, CircleAvg2 ... CircleAvgn).
Again, the latter number in this average gray value sequence is subtracted into previous number (except the last one), obtained
One new sequence:
(CircleAvg2-CircleAvg1, CircleAvg3-CircleAvg2 ... CircleAvgn-
CircleAvgn-1)
Wherein each value, all representing respective radius R1 ... Rn has a score.
Finally, finding out scoring of the highest score as the center in the new sequence, the corresponding radius of highest score is made
For the optimum radius at the center, specific formula is as follows:
Scorecenter=max (CircleAvg2-CircleAvg1, CircleAvg3-CircleAvg2,
...CircleAvgn-CircleAvgn-1)
(4.2) center is wrapped as shown in Figure 7 and Figure 8 in the method for progressive step length searching scoring of series of points around it
Include step:
(a) center scoring is calculated, as initial score.Center is the estimation center of circle under original state, and initial score is
Scorecenter。
(b) eight neighborhood for being separated by 2 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin scoring ScorecenterCompare, if maximum value is greater than initial score, then point corresponding to maximum value is returned as new center
Step (a) is returned, is otherwise continued.
It scores specifically, calculating the eight neighborhood for being separated by 2 pixels with center:
(estCenterX-2, estCenterY-2), (estCenterX, estCenterY-2),
(estCenterX+2, estCenterY-2), (estCenterX-2, estCenterY),
(estCenterX+2, estCenterY), (estCenterX-2, estCenterY+2),
(estCenterX, estCenterY+2), (estCenterX+2, estCenterY+2).
Using this eight points as the center of circle, the scoring of each point is found out with the methods of marking, takes the maximum value in scoring later:
Score2-maxWith the initial score ScorecenterIt compares.
If Score2-max> Scorecenter, then using the corresponding center of circle as new center return step (a).
If Score2-max< Scorecenter, then continuing below step.
(c) eight neighborhood for being separated by 4 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue.
It scores specifically, calculating the eight neighborhood for being separated by 4 pixels with center:
(estCenterX-4, estCenterY-4), (estCenterX, estCenterY-4),
(estCenterX+4, estCenterY-4), (estCenterX-4, estCenterY),
(estCenterX+4, estCenterY), (estCenterX-4, estCenterY+4),
(estCenterX, estCenterY+4), (estCenterX+4, estCenterY+4).
Using this eight points as the center of circle, the scoring of each point is found out with the methods of marking, takes the maximum value in scoring later
Score4-maxWith the initial score ScorecenterIt compares.
If Score4-max> Scorecenter, then using the corresponding center of circle as new center return step (a).
If Score4-max< Scorecenter, then continuing below step.
(d) eight neighborhood for being separated by 6 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue.
It scores specifically, calculating the eight neighborhood for being separated by 6 pixels with center:
(estCenterX-6, estCenterY-6), (estCenterX, estCenterY-6),
(estCenterX+6, estCenterY-6), (estCenterX-6, estCenterY),
(estCenterX+6, estCenterY), (estCenterX-6, estCenterY+6),
(estCenterX, estCenterY+6), (estCenterX+6, estCenterY+6).
Using this eight points as the center of circle, the scoring of each point is found out with the methods of marking, takes the maximum value in scoring later
Score6-maxWith the initial score ScorecenterIt compares.
If Score6-max> Scorecenter, then using the corresponding center of circle as new center return step (a).
If Score6-max< Scorecenter, then continuing below step.
(e) eight neighborhood for being separated by 8 pixels with center is calculated to score；Take the eight neighborhood score in maximum value and it is described just
Begin to score and compare, if maximum value is greater than initial score, then by point corresponding to maximum value as new center return step
(a), otherwise continue.
It scores specifically, calculating the eight neighborhood for being separated by 8 pixels with center:
(estCenterX-8, estCenterY-8), (estCenterX, estCenterY-8),
(estCenterX+8, estCenterY-8), (estCenterX-8, estCenterY),
(estCenterX+8, estCenterY), (estCenterX-8, estCenterY+8),
(estCenterX, estCenterY+8), (estCenterX+8, estCenterY+8).
Using this eight points as the center of circle, the scoring of each point is found out with the methods of marking, takes the maximum value in scoring later
Score8-maxWith the initial score ScorecenterIt compares.
If Score8-max>Scorecenter, then using the corresponding center of circle as new center return step (a).
If Score8-max<Scorecenter, then continuing below step.
(f) if the scoring of above-mentioned all the points is respectively less than initial score, then the center that iteration is gone out is the accurate center of circle, and
And using the optimum radius at the center as precise radius.
Specifically, if above-mentioned all the points are both less than Score as the scoring Score in the center of circlecenter, it is considered that iteration
The center estimated is the accurate center of circle, and using the optimum radius at the center as final radius.
(5) the accurate center of circle of iris and precise radius are sent to client.
For the realization for cooperating the method for the present invention, it is based on identical inventive concept, it is shown in Figure 9, the present invention also provides
The system of human eye iris in a kind of detection picture, including it is image collection module, image conversion module, estimation computing module, accurate
Computing module and output module；
Described image obtains module, obtains eyes image using client camera；
The eyes image is converted gray level image by described image conversion module；
The estimation computing module is filtered calculating and rectangular block gray scale Data-Statistics to the gray level image, obtains rainbow
The estimation center of circle of film and estimation radius；
It is described to accurately calculate module, according to the estimation center of circle and estimation radius, utilize the side in the progressive step-length resetting center of circle
Method iteratively solves the accurate center of circle and the precise radius of iris；
The accurate center of circle of iris and precise radius are sent to client by the output module.
The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention.The technology of the industry
Personnel are it should be appreciated that the present invention is not limited to the above embodiments, and the above embodiments and description only describe this
The principle of invention, without departing from the spirit and scope of the present invention, various changes and improvements may be made to the invention, these changes
Change and improvement all fall within the protetion scope of the claimed invention.This reality invent claimed range by appended claims and
Its equivalent thereof.
Claims (9)
1. a kind of method of human eye iris in detection picture, which is characterized in that comprising steps of
(1) eyes image is obtained using client camera；
(2) gray level image is converted by the eyes image；
(3) calculating and rectangular block gray scale Data-Statistics are filtered to the gray level image, obtain the estimation center of circle and the estimation of iris
Radius；
In the step (3), the rectangular block gray scale Data-Statistics are comprising steps of create horizontal rectangular in the gray level image
Block；From left to right move the horizontal rectangular block, all pixels gray scale when every one pixel of movement in statistics horizontal rectangular block
The center of the sum of value and horizontal rectangular block；The smallest horizontal rectangular block of the sum of gray value is found out, its center is made
For the horizontal position of pupil；Vertical rectangular block is created, by the vertical rectangular block in the pixel of that column of the horizontal position
Vertical sliding, when every one pixel of movement, count the sum of all pixels gray value in vertical rectangular block and vertical rectangular block
Center；The smallest vertical rectangular block of the sum of gray value is found out, using its center as the vertical position of pupil；Statistics institute
The gray value for stating that column pixel of horizontal position forms grey level histogram；By the corresponding gray value in the lowest point of the grey level histogram
Comparison threshold value is obtained multiplied by pre-determined factor；It counts gray value in that column pixel of the horizontal position and is less than the comparison threshold value
Pixel；By the number of institute's statistical pixel divided by 2, the pupil radium estimated；The horizontal position of the pupil and vertical position
As the estimation center of circle of iris, estimation radius of the pupil radium as iris；
(4) it according to the estimation center of circle and estimation radius, is iteratively solved using the method in the progressive step-length resetting center of circle, obtains iris
The accurate center of circle and precise radius；
(5) the accurate center of circle of iris and precise radius are sent to client.
2. the method for human eye iris in a kind of detection picture according to claim 1, which is characterized in that the step (1)
In, the acquisition methods of the eyes image are to detect the key point that face obtains detection face by machine learning, extract eye
Key point constitutes eyes image；Or it by user's interaction, clicks screen and obtains eyes image.
3. the method for human eye iris in a kind of detection picture according to claim 1, which is characterized in that the step (2)
In, the eyes image is RGB image, converts ash for eyes image using the value in tri- channels R, G and B in RGB image
Spend image.
4. the method for human eye iris in a kind of detection picture according to claim 1, which is characterized in that the step (3)
In, centered on the filtering calculates comprising steps of setting some pixel in one-row pixels in the gray level image；Find out institute
State the pixel near central horizontal direction；Calculate the gray value at the center and the pixel nearby；Gray value is ascending
Sequence；The gray value at the center is replaced with maximum or secondary big gray value.
5. the method for human eye iris in a kind of detection picture according to claim 1, which is characterized in that the horizontal rectangular
Frame height degree is eyes image height, and width is a quarter of eyes image width；The vertical rectangle frame height is eye figure
The a quarter of image height degree, width are eyes image width；The coefficient value is 1.4.
6. a kind of method for detecting human eye iris in picture according to claim 1 or 5, which is characterized in that the step
(4) in, the method iterative solution in the progressive step-length resetting center of circle, comprising steps of first using the estimation center of circle as the initial center of circle, and
Find out the scoring and respective radius in the initial center of circle；Then centered on the initial center of circle, according to progressive step length searching its
The scoring and respective radius of surrounding series of points；If the scoring for being scored above the initial center of circle of some point, just using the point as new
Center；To new center still with progressive step length searching around it series of points scoring and respective radius, carry out a new round
Iteration；After iteration, if the scoring all put is respectively less than the scoring at center, using the center as the accurate center of circle, and should
The respective radius at center is as precise radius.
7. the method for human eye iris in a kind of detection picture according to claim 6, which is characterized in that the methods of marking
Comprising steps of setting initial radium, the sum of the gray value on part-circular periphery is counted according to the center；Later by the gray value
The sum of divided by the number of pixels on part-circular periphery, obtain circumference average gray value；Certain Radius is incremented to since initial radium,
Form radii sequence；The corresponding circumference average gray value of all radiuses in radii sequence is sought, average gray value sequence is obtained；By institute
It states the latter value in average gray value sequence and subtracts previous value, obtain new sequence, each value, all generation in the new sequence
The score of table respective radius；Find out scoring of the highest score in the new sequence as the center, highest score corresponding half
Best respective radius of the diameter as the center.
8. the method for human eye iris in a kind of detection picture according to claim 7, which is characterized in that center with progressive
The method of step length searching scoring of series of points around it comprising steps of
(a) center scoring is calculated, as initial score；
(b) eight neighborhood for being separated by 2 pixels with center is calculated to score；The maximum value in the eight neighborhood scoring is taken initially to comment with described
Divide and compare, if maximum value is greater than initial score, then point corresponding to maximum value is no as new center return step (a)
Then continue；
(c) eight neighborhood for being separated by 4 pixels with center is calculated to score；The maximum value in the eight neighborhood scoring is taken initially to comment with described
Divide and compare, if maximum value is greater than initial score, then point corresponding to maximum value is no as new center return step (a)
Then continue；
(d) eight neighborhood for being separated by 6 pixels with center is calculated to score；The maximum value in the eight neighborhood scoring is taken initially to comment with described
Divide and compare, if maximum value is greater than initial score, then point corresponding to maximum value is no as new center return step (a)
Then continue；
(e) eight neighborhood for being separated by 8 pixels with center is calculated to score；The maximum value in the eight neighborhood scoring is taken initially to comment with described
Divide and compare, if maximum value is greater than initial score, then point corresponding to maximum value is no as new center return step (a)
Then continue；
If (f) scoring of above-mentioned all the points is respectively less than initial score, then the center that iteration is gone out is the accurate center of circle, and will
The optimum radius at the center is as precise radius.
9. the system of human eye iris in a kind of detection picture of any one the method for realizing claim 1-8,
It is characterized in that, including image collection module, image conversion module, estimates computing module, accurately calculates module and output module；
Described image obtains module, obtains eyes image using client camera；
The eyes image is converted gray level image by described image conversion module；
The estimation computing module is filtered calculating and rectangular block gray scale Data-Statistics to the gray level image, obtains iris
Estimate the center of circle and estimation radius；
It is described to accurately calculate module, according to the estimation center of circle and estimation radius, changed using the method in the progressive step-length resetting center of circle
In generation, solves the accurate center of circle and the precise radius of iris；
The accurate center of circle of iris and precise radius are sent to client by the output module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510984804.4A CN105389574B (en) | 2015-12-25 | 2015-12-25 | The method and system of human eye iris in a kind of detection picture |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510984804.4A CN105389574B (en) | 2015-12-25 | 2015-12-25 | The method and system of human eye iris in a kind of detection picture |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105389574A CN105389574A (en) | 2016-03-09 |
| CN105389574B true CN105389574B (en) | 2019-03-22 |
Family
ID=55421846
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510984804.4A Active CN105389574B (en) | 2015-12-25 | 2015-12-25 | The method and system of human eye iris in a kind of detection picture |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105389574B (en) |
Families Citing this family (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106127830A (en) * | 2016-09-06 | 2016-11-16 | 成都通甲优博科技有限责任公司 | The making of a kind of virtual beauty pupil pupil sheet and display packing |
| CN106529423A (en) * | 2016-10-20 | 2017-03-22 | 鹄誉医疗科技（上海）有限公司 | Energy bar human-eye distance measurement method based on Android system |
| CN106491073A (en) * | 2016-10-20 | 2017-03-15 | 鹄誉医疗科技（上海）有限公司 | A kind of human eye pupil detection method of quick high robust |
| CN106530229B (en) * | 2016-11-07 | 2019-05-21 | 成都通甲优博科技有限责任公司 | The production method and display methods of virtualization processing virtual beauty pupil pupil piece afterwards |
| CN107451546A (en) | 2017-07-14 | 2017-12-08 | 广东欧珀移动通信有限公司 | Iris identification method and related product |
| CN108288248A (en) * | 2018-01-02 | 2018-07-17 | 腾讯数码（天津）有限公司 | A kind of eyes image fusion method and its equipment, storage medium, terminal |
| CN109902569B (en) * | 2019-01-23 | 2021-09-17 | 上海思立微电子科技有限公司 | Fingerprint image conversion method and device and fingerprint identification method |
| CN112434675B (en) * | 2021-01-26 | 2021-04-09 | 西南石油大学 | Pupil positioning method for global self-adaptive optimization parameters |
| CN112906596A (en) * | 2021-03-03 | 2021-06-04 | 杭州电子科技大学 | Iris center positioning method based on low-resolution optical camera |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1794262A (en) * | 2005-12-29 | 2006-06-28 | 上海交通大学 | Method of fast and effective cutting apart vivi iris |
| CN1885314A (en) * | 2006-07-11 | 2006-12-27 | 电子科技大学 | Pre-processing method for iris image |
| CN101201893A (en) * | 2006-09-30 | 2008-06-18 | 电子科技大学中山学院 | Iris recognizing preprocessing method based on grey level information |
| CN101266645A (en) * | 2008-01-24 | 2008-09-17 | 电子科技大学中山学院 | Iris positioning method based on multi-resolutions analysis |
-
2015
- 2015-12-25 CN CN201510984804.4A patent/CN105389574B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1794262A (en) * | 2005-12-29 | 2006-06-28 | 上海交通大学 | Method of fast and effective cutting apart vivi iris |
| CN1885314A (en) * | 2006-07-11 | 2006-12-27 | 电子科技大学 | Pre-processing method for iris image |
| CN101201893A (en) * | 2006-09-30 | 2008-06-18 | 电子科技大学中山学院 | Iris recognizing preprocessing method based on grey level information |
| CN101266645A (en) * | 2008-01-24 | 2008-09-17 | 电子科技大学中山学院 | Iris positioning method based on multi-resolutions analysis |
Non-Patent Citations (1)
| Title |
|---|
| 虹膜定位算法研究;李庆嵘等;《电子科技大学学报》;20020228;第31卷(第1期);第7-8页 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105389574A (en) | 2016-03-09 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105389574B (en) | The method and system of human eye iris in a kind of detection picture | |
| CN107133969B (en) | A kind of mobile platform moving target detecting method based on background back projection | |
| CN109614922B (en) | Dynamic and static gesture recognition method and system | |
| CN110210360B (en) | Rope skipping counting method based on video image target recognition | |
| CN103927016B (en) | Real-time three-dimensional double-hand gesture recognition method and system based on binocular vision | |
| WO2019223069A1 (en) | Histogram-based iris image enhancement method, apparatus and device, and storage medium | |
| CN110221699B (en) | Eye movement behavior identification method of front-facing camera video source | |
| CN103345644B (en) | The object detection method of on-line training and device | |
| CN103955949B (en) | Moving target detecting method based on Mean-shift algorithm | |
| CN109918971B (en) | Method and device for detecting number of people in monitoring video | |
| CN105117705B (en) | A kind of iris image quality tandem type evaluation method | |
| CN103902958A (en) | Method for face recognition | |
| CN103136519A (en) | Sight tracking and positioning method based on iris recognition | |
| CN108596041A (en) | A kind of human face in-vivo detection method based on video | |
| CN106056064A (en) | Face recognition method and face recognition device | |
| CN106911921A (en) | The infrared touch-control of projecting apparatus and Atomatic focusing method based on single camera | |
| CN104281839A (en) | Body posture identification method and device | |
| CN107169427B (en) | Face recognition method and device suitable for psychology | |
| CN103218615B (en) | Face judgment method | |
| CN110163567A (en) | Classroom roll calling system based on multitask concatenated convolutional neural network | |
| CN109360179A (en) | A kind of image interfusion method, device and readable storage medium storing program for executing | |
| CN104008364A (en) | Face recognition method | |
| CN107895157A (en) | A kind of pinpoint method in low-resolution image iris center | |
| CN110414558A (en) | Characteristic point matching method based on event camera | |
| CN111259757B (en) | Living body identification method, device and equipment based on image |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CP02 | Change in the address of a patent holder |
Address after: 610041 the 13 floor of No. 1 middle Tianfu Avenue, No. 1268, high-tech zone, Chengdu, Sichuan. Patentee after: Chengdu PinGuo Digital Entertainment Ltd. Address before: 3, 13 floor, E zone, Tianfu Software Park, Chengdu hi tech Zone, Sichuan, 610000 Patentee before: Chengdu PinGuo Digital Entertainment Ltd. |
|
| CP02 | Change in the address of a patent holder |